Chapter Download THE GREEN AND VIRTUAL DATA CENTER Chapter 8: Data Storage — Disk, Tape, Optical, and Memory By Greg Schulz Sponsored By: Chap8.fm Page 205 Thursday, October 23, 2008 5:28 PM Chapter 8 Data Storage—Disk, Tape, Optical, and Memory I can’t remember where I stored that. In this chapter you will learn that: An expanding data footprint results in increased management costs and complexity. Data storage management is a growing concern from a green stand- point. There are many aspects of storage virtualization that can help address green challenges. Demand to store more data for longer periods of time is driving the need for more data storage capacity, which in turn drives energy consump- tion and consequent cooling demands. This chapter looks at various data storage technologies and techniques used to support data growth in an eco- nomical and environmentally friendly manner. These technologies also aid in sustaining business growth while building on infrastructure resource management functions, including data protection, business continuance and disaster recovery (BC/DR), storage allocation, data movement, and migration, along with server, storage, and networking virtualization topics. Although this chapter focuses on external direct attached and networked storage (either networked attached storage or a storage area network), the principles, techniques, and technologies also apply to internal dedicated storage. The importance of this chapter is to understand the need to sup- port and store more data using various techniques and technologies to enable more cost-effective and environmentally as well as energy-friendly data growth. 205 Chap8.fm Page 206 Thursday, October 23, 2008 5:28 PM 206 The Green and Virtual Data Center 8.1 Data Storage Trends, Challenges, and Issues After facilities cooling for all IT equipment and server energy usage, exter- nal data storage has the next largest impact on power, cooling, floor space, and environmental (PCFE) considerations in most environments. In addi- tion to being one of the large users of electrical power and floor space, with corresponding environmental impact, the amount of data being stored and the size of its the data footprint continue to expand. Though more data can be stored in the same or smaller physical foot- print than in the past, thus requiring less power and cooling, data growth rates necessary to sustain business growth, enhanced IT service delivery, and new applications are placing continued demands on available PCFE resources. A key driver for the increase in demand for data storage is that more data is being generated and stored for longer periods of time as well as more copies of data in multiple locations. This trend toward increasing data stor- age will likely not slow anytime soon for organizations of all sizes. The popularity of rich media and Internet-based applications has resulted in explosive growth of unstructured file data that requires new and more scalable storage solutions. Applications such as video pre- and post- production processing, animation rendering, on-demand video and audio, social networking websites, and digitalization of data from cell phones, per- sonal digital assistants (PDAs) and other sources have increased burdens on storage performance and capacity. Unstructured data includes spreadsheets, PowerPoint presentations, slide decks, Adobe PDF and Microsoft Word documents, Web pages, and video and audio JPEG, MP3, and MP4 files. The diversity of rich media and Internet applications ranges from many small files with various access patterns to more traditional large video stream access. Consequently, storage systems, in order to scale with stability to sup- port Internet and Web 2.0 applications, will need to support variable per- formance characteristics from small random access of meta-data or individual files to larger streaming video sequences. Data growth rates range from the low double digits to high double or triple digits as more data is generated, more copies of data are made and, more data is stored for longer periods of time. While structured data in the form of databases continues to grow, for most environments and applications, it is semistructured email data and Chap8.fm Page 207 Thursday, October 23, 2008 5:28 PM Data Storage Trends, Challenges, and Issues 207 Figure 8.1 Expanding Data Footprint due to Data Proliferation and Copies Being Retained unstructured file data that creates the biggest data footprint impact and sub- sequent bottlenecks. Unstructured data has varying input/output (I/O) characteristics that change over time, such as data that starts out with a lot of activity, then goes idle for a time before extensive reads, as in the case of a video or audio file becoming popular on a media, entertainment, social net- working, or company-sponsored website. As another example, usually, when a development or research project is completed, the data or intellec- tual property is archived or migrated to lower-cost, lower-performance bulk storage until it is needed again for further research or sequel projects. The data footprint is the total data storage needed to support applica- tion and information needs. Your data footprint may, in fact, be larger than the actual amount of data storage you have, or you may have more aggregated data storage capacity than actual data. A general approach to determine your data footprint is simply to add up all of your online, near- line, and offline data storage (disk and tape) capacity. For example, con- sider all the data being stored at home on personal computers and laptops, PDAs, digital cameras and video recorders, TiVo sets and DVRs, USB fixed and removable disk drives, among other media that support various data and information needs. Digital households, that is, homes with one or more computers and other electronic equipment, may have from 500 GB to over 1 TB of data. These homes’ storage needs will continue to grow. The importance of understanding digital data growth needs for homes is to be able to put into Chap8.fm Page 208 Thursday, October 23, 2008 5:28 PM 208 The Green and Virtual Data Center scale the amount of data that needs to be stored in IT data centers to sup- port existing and emerging applications and services. Suppose that a business has 20 TB of data storage space that is allocated and being used for databases, email, home directories, shared documents, engineering documents, financial, and other data in different formats, both structured and unstructured. For these 20 TB of data, the storage space is probably not 100% used; database tables may be sparsely allocated, and there is likely duplicate data in email and shared document folders. How- ever, to keep the example straightforward, assume that of the 20 TB, two complete copies are required for BC/DR purposes, and 10 TB are dupli- cated to three different areas on a regular basis for application testing, train- ing, and business analysis and reporting. See Figure 8.1. The overall data footprint is the total amount of data, including all copies plus the additional storage required to support that data, such as extra disks for redundant array of independent disks (RAID) protection or remote mirroring. In this overly simplified example, the data footprint and subsequent storage requirement amount to several times the 20 TB of data. And the larger the data footprint, the more data storage capacity and perfor- mance bandwidth are needed and that have to be powered, cooled, and housed in a rack or cabinet on a floor somewhere. Costs associated with supporting an increased data footprint include: Data storage hardware and management software tools acquisition Associated networking or I/O connectivity hardware and services Recurring maintenance and software renewal fees Facilities fees for floor space, power, and cooling Physical and logical security of data and IT technology assets Data protection for high availability and BC/DR, including backup, replication, and archiving It is debatable how much energy in a typical data center is actually con- sumed by storage (internal to servers and external) as well as how much data is active or inactive. The major power draws for common storage systems are usually spinning hard disk drives (HDDs) and their enclosures, which account for, on average, 66–75%; controllers and related I/O connectivity Chap8.fm Page 209 Thursday, October 23, 2008 5:28 PM Addressing PCFE Storage Issues 209 components generally account for most of the balance of electrical power consumption. Consequently, data storage is an important area for energy optimization and efficiency improvements. One approach to reducing your footprint is simply to stop spending and put a cap on growth. For most environments, freezing growth is a bit draconian, but it is an option. A better approach is to do more with what you have or do more with less—that is, enable growth via consolidation and optimization to the point where further consolidation and optimization become self-defeating. 8.2 Addressing PCFE Storage Issues There are many approaches to addressing PCFE issues associated with storage, from using faster, more energy efficient storage that performs more work with less energy, to powering down storage that is supporting inactive data, such as backup or archive data, when it is not in use. While adaptive and intelligent power management techniques are increasingly being found in servers and workstations, power management for storage has lagged behind. General steps to doing more with your storage-related resources with- out negatively impacting application service availability, capacity, or perfor- mance include: Assess and gain insight as to what you have and how it is being used. Develop a strategy and plan (near-term and long-term) for deploy- ment. Use energy-effective data storage solutions (both hardware and soft- ware). Optimize data and storage management functions. Shift usage habits to allocate and use storage more effectively.